drza wrote:[...]
Excellent portability & scalability: http://hoopslab.rotowire.com/post/162857973161/lebron-james-impact-diminishing-returns-compared
The ability to be almost purely additive with no downside: http://hoopslab.rotowire.com/post/162035208506/individual-offense-is-not-more-important-than
And is historically scarce: http://hoopslab.rotowire.com/post/162019505846/individual-offense-is-not-more-important-than
[...]
I'm glad you posted these links since I had no idea you were a writer. I'll have to read through your past content at some point. I was actually about to make a thread on the matter -- that is, on the additivity of defense -- since I felt as essentially a sub-topic it didn't get all the focus it deserved. Not to become off-topic, but as a concept that has influenced the voting in this thread, I was wondering if you don't mind answering a few questions:
- 1. I wish there was some conceptual explanation for why defense might be more additive than offense. Looking at the example you provide:
Let’s go back to the original example I used, of a +6 offensive player vs a +5 defensive player... Suppose your weakest link is a +3 offensive player. If you replace him with the +6 offense player, who let’s say plays the same position, how much will the team’s offense improve (in rough terms)? Will it improve by +6? Will it improve by +3? Of course it’s a complicated question, but the answer is very likely no to either number... But in the same type of example, if you replace a +2 defensive player by a +5…you’ll see a LOT more of the value transfer through.
I can't escape how replacing every instance of "defensive player" with "offensive player" yields an equally sensible statement. Thus, I don't feel this example illustrates why defensive impact might be more additive than offensive impact. Rather, I think it's a general statement about impact. Setting aside any statistical evidence, why do you think defensive impact is more additive than offensive impact? How would you frame such an argument when defensive skills can often be interpreted as counters to offensive skills, and seemingly, the two are mirrors?
2. Do you think it's possible that the empirical additivity of defensive impact, compared to offensive impact, is because greater collections of offensive talent have been assembled? That is, perhaps teams are more likely assemble a (projected) +15 offense than a (projected) +15 defense, and thus the diminishing returns to offensive impact is present at a greater degree? If I had my computer working, I would be trying to answer this myself, but currently I cannot.